MPC实用化问题处理及在车辆ACC中的应用

Processing of MPC practical problems and its application to vehicular adaptive cruise control systems

为实现多目标协调式自适应巡航控制(ACC)系统的实车应用,分析并解决了模型预测控制(MPC)理论实用化过程的弱鲁棒性、非可行解和高计算复杂度问题。采用反馈校正法补偿跟车模型的预测误差,改善模型对跟车系统状态的预测精度;再利用约束管理法,修正MPC代价函数,松弛其输入输出(I/O)硬约束;基于变量集结法,降低待优化变量的维数,缩减MPC优化问题的规模。以某重型卡车为对象的ACC仿真表明:该方法可有效提高ACC对模型失配的鲁棒性,避免因过大跟踪误差造成的控制律非可行解,提高MPC计算效率的同时不影响其控制最优性。

Low robustness,computing infeasibility and high computational complexity in model predictive control（MPC） theory were analyzed to apply the multi-objective coordinated adaptive cruise control（ACC） system to vehicle products.The feedback correction method was used to compensate the predictive car-following model errors and improve the predictive system state precision.With the constraint management method used to revise the cost function and soften the optimization problem input/output（I/O） constraints,a variable aggregation strategy was used to decrease the optimized variable dimension,thus reducing the scale of MPC optimization problem.Series of vehicular ACC simulations with a heavy-duty truck plant show that the methods enhance vehicular ACC＇s robustness to model mismatch,avoid computing infeasibility of control law due to large tracking errors and increase the MPC computing efficiency while not affecting its control optimality.